In today’s post I’d like to explain how I used the following formula to help merge multiple columned data into a different columned arrangement as the purpose of the result was to export the original data set into a CSV file for import into NetSuite.

Here’s the formula we will dissect:

=ARRAYFORMULA(SPLIT(ARRAYFORMULA(SUBSTITUTE(TRANSPOSE(SPLIT(JOIN(" ",QUERY(DataConcat!A:L,,ROWS(DataConcat!A:A)))," ")),"~"," ")),";",TRUE,FALSE))

The problem

I had exported the following salary data from our budgeting software which contained how much each staff person would be receiving each month throughout the financial year.

An example of the raw data looked something like this:

Employee	Jan-2020	Feb-2020
123 Smith, Bob	2,000	2,200
456 Doe, Jane	3,000	3,300
789 Doe, John		200

The original spreadsheet had all months of the year and is labelled as Data:

I had to import this data into NetSuite using the CSV Import process, which required changing the structure a little bit, so that the resulting CSV looked like so:

Employee ID	Employee Name	From Date	To Date	Amount
123	Smith, Bob	1/01/2020	31/01/2020	2,000
123	Smith, Bob	1/02/2020	29/02/2020	2,200
456	Doe, Jane	1/01/2020	31/01/2020	3,000
456	Doe, Jane	1/02/2020	29/02/2020	3,300
789	Doe, John	1/02/2020	29/02/2020	200

As you can see with the requirements of the task, we need to be able to expand the data set to rows (as it’s currently laid out in columns), then merge the data back into one columnar set. Once we have achieved this we then use the handy File > Download > Comma-separated values (current sheet) option in Google Sheets.

Solution: Step 1 – Expansion

The first rule when manipulating data is to leave the original data set intact.

Therefore, we create a fresh new sheet and label this whatever name is appropriate for your needs. In this example we will name it appropriate to what we are doing: DataExpand.

In our expansion we want to achieve all necessary CSV fields required. This means we need to fetch the ID number from the original Employee field, and to copy the remaining data into another field for the Employee Name. Then we split the header column in our original sheet to a data range. Followed lastly with the amount in each date range.

Our first sheet, should look a little something like this:

Here are the details as displayed in the photo above:

• Cells $B$1 and $G$1 contained the =COLUMN() reference to each of the months on the original data set. This just helps with creating an easy index function that uses that reference which can make it easier copying the same columnar set 12 times.
• Row 1 this contains the following formula to extract the ID number from the Employee field. Here we check to make sure we have data in the original Data sheet, if we do we want to perform a regular expression where we just want all the digits, therefore, we use the regular expression of d+:

=if(len(Data!A2),REGEXEXTRACT(Data!A2,"d+"),"")

• From $B$2 down we then use another regex formula to extract those digits and replace them with nothing. We also want to check if the employee received any salaries or wages, if they didn’t we don’t want to add them into the data set.

=if(index(Data!$1:$1000,row(),B$1),trim(regexreplace(Data!$A2,"d+","")),"")

• From $C$2 down we just insert the ID that is needed for each row if we have a confirmed value in the previous column.

=if(len(B2),$A2,"")

• For $D$2 down if we have a value in $B$2 then we get the value of the salaries for that respective month. We use the handy index function here to get that value:

=if(len(B2),index(Data!$1:$1000,row(),B$1),"")

• For $E$2 down if we have a value in $B$2 then we get the column heading of the respective month using, once again, the index function:

=if(len(B2),index(Data!$1:$1000,1,B$1),"")

• Lastly, for $F$2 down we get the value in the previous cell, if something exists, and calculate what the last day of the month will be for that cell:

=if(len(E2),date(year(E2),month(E2)+1,0),"")

As we have been mindful of the formulas used for each budget month, we now just need to copy cells $B$2:$F$1 and then paste this 12 times across. We could make it even easier by, after our first paste in cells $G$2:$K$4 changing cell $G$1 to be a formula: =$B$1 + 1:

Then we copy this newly pasted range with formula edits across 10 more times.

Solution: Step 2 – Concatenation

The purpose of the next sheet is to zip up our columnar collections. This is a relatively simple step, however, there a couple of things to be wary of:

1. Use TEXTJOIN when zipping up the columnar data rather than the JOIN or CONCATENATE functions. The primary reason for this is where there is no data in a cell TEXTJOIN creates the blank join, whereas the other functions neglect it. As an example, if we had the following table:

	A	B	C
1	Missing	Puzzle	Piece
2	Missing		Piece

When using =JOIN(",",A1:C1) it produces the result Missing,Puzzle,Piece for the first row and =JOIN(",",A2:C2) will produce Missing,Piece for the second row.

Whereas using =TEXTJOIN(",",FALSE,A1:C1) produces the result Missing,Puzzle,Piece for the first row and =TEXTJOIN(",", FALSE, A2:C2) will produce Missing,,Piece for the second row.

As the TEXTJOIN function allows for the flexibility of the concatenation of your columnar data it’s import the data remain as it is represented elsewhere if there happens to be blank cells in your data.

1. The other important aspect with this formula is that we want to replace the space characters within our cells as the space character will be used to join and split data back into columns again later. Therefore, it’s important at this step (if your data also contains spaces) to replace the space character with an oblique character that is not used anywhere in your data set.

I am going to replace my space characters within my data to the tilde character ~.

Therefore, I have created a new sheet, which I’ve labelled DataConcat and cell $A$1 has the formula:

=if(len(DataExpand!B2),SUBSTITUTE(TEXTJOIN(";",FALSE,DataExpand!B2:F2)," ","~"),"")

Be mindful when transferring the data across to adjacent columns that you don’t copy paste or drag across the formula, as each column here needs to reference the collection in DataExpand. For example, in the next adjacent cell $B$1 the formula is:

=if(len(DataExpand!G2),substitute(textjoin(";",FALSE,DataExpand!G2:K2)," ","~"),"")

You will need to do this 12 times for each month.

Solution: Step 3 – Merge

So here’s where we merge all the columns from the DataConcat sheet in Step 2 above into one big column collection.

=ARRAYFORMULA(SPLIT(ARRAYFORMULA(SUBSTITUTE(TRANSPOSE(SPLIT(JOIN(" ",QUERY(DataConcat!A:L,,ROWS(DataConcat!A:A)))," ")),"~"," ")),";",TRUE,FALSE))

Wow, what a massive formula!

As with mathematics, we have to start with the inner most functions and work our way out. So let’s break up each component within this formula and explain what’s happening:

QUERY function

QUERY(DataConcat!A:L,,ROWS(DataConcat!A:A))

Here we use the QUERY function to capture the data we want to operate on (the first parameter), being all the columns in the DataConcat sheet, and we then want to combine these cells into one row. We achieve this by setting the number of headers (the third parameter) to the maximum number of rows in the DataConcat sheet. This means every row in the DataConcat sheet will be zipped up into one row!

The way the QUERY function zips everything up is by applying the space character to each new cell. This was why it was important for us to remove the space characters from our own data set before we applied this function.

If you’d like to know more about the QUERY function I wrote up a more detailed post here explaining this type of use.

MERGE – Using JOIN, SPLIT & TRANSPOSE functions

TRANSPOSE(SPLIT(JOIN(" ", QUERY(...)), " "))

The combination of these three functions is to merge the single row returned by the QUERY function into one column.

First we JOIN the rows together using the special space character which helps to delimit each individual row. Then we SPLIT by the space character which produces the rows across individual cells horizontally. Finally, the TRANSPOSE function translates the horizontal cells into vertical formation – a single column.

Unfortunately the JOIN function does have a character length limitation of 50,000 characters. You may need to break up your requirements or look at other means if this ends up being an impediment.

EXPANSION – Using ARRAYFORMULA

The last step is expand our singular column so that it can be represented as needed for CSV export. If we’re expanding data sets we need to use the ARRAYFORMULA function, as demonstrated here:

ARRAYFORMULA(SPLIT(SUBSTITUTE(transpose(...), "~", " "), ";", TRUE, FALSE))

The ARRAYFORMULA provides us with the ability to loop through each of the cells in our data set and to apply individual functions to them. In our case we want to:

1. SUBSTITUTE back the oblique tilde ~ character with its original space character.
2. SPLIT back the data we concatenated in the DataConcat sheet. However, we need to be mindful there may be instances where we have blank cells, and we want the SPLIT function to return the blank cells, therefore, we need to change our fourth parameter in the SPLIT function to FALSE (default is TRUE). By switching this to FALSE it turns consecutive delimiters like ;; into three separate cells, rather than just one cell.

With these functions all combined they produce the wonderful result intended, like so:

Conclusion

Within this post we were able to learn how we can extrapolate data from one form to another using several intermediary steps and complex Google Sheet functions.

Hopefully this post has been helpful with learning:

• How to organise your steps by leaving the original data set (use multiple sheets if needed).
• How to use the QUERY function and especially its third headers parameter.
• Why you would use the TEXTJOIN function over JOIN.
• The limitations of the JOIN function.
• How to expand your data set using ARRAYFORMULA with any function when needing to loop through individual cells in the result.
• How to set the SPLIT function’s fourth parameter so that consecutive empty cells aren’t treated as one.